Many diseases of aging including AD (Alzheimer's disease) and T2D (Type 2 diabetes) are strongly associated with common risk factors, suggesting that there may be shared aging mechanisms underlying these diseases, with the scope to identify common cellular targets for therapy. In the present study we have examined the insulin-like signalling properties of an experimental AD 8-hydroxyquinoline drug known as CQ (clioquinol). The IIS [insulin/IGF-1 (insulin-like growth factor-1) signalling] kinase Akt/PKB (protein kinase B) inhibits the transcription factor FOXO1a (forkhead box O1a) by phosphorylating it on residues that trigger its exit from the nucleus. In HEK (human embryonic kidney)-293 cells, we found that CQ treatment induces similar responses. A key transcriptional response to IIS is the inhibition of hepatic gluconeogenic gene expression, and, in rat liver cells, CQ represses expression of the key gluconeogenic regulatory enzymes PEPCK (phosphoenolpyruvate carboxykinase) and G6Pase (glucose-6-phosphatase). The effects on FOXO1a and gluconeogenic gene expression require the presence of Zn2+ ions, reminiscent of much earlier studies examining diabetogenic properties of 8-hydroxyquinolines. Comparative investigation of the signalling properties of a panel of these compounds demonstrates that CQ alone exhibits FOXO1a regulation without diabetogenicity. Our results suggest that Zn2+-dependent regulation of FOXOs and gluconeogenesis may contribute to the therapeutic properties of this drug. Further investigation of this signalling response might illuminate novel pharmacological strategies for the treatment of age-related diseases.
- Alzheimer's disease
- forkhead box O1a (FOXO1a)
- Type 2 diabetes
Besides its role in glucose homoeostasis, IIS [insulin/IGF-1 (insulin-like growth factor-1) signalling] components are increasingly understood to control complex relationships between diet and longevity in a wide variety of species. In animal models for example, lifelong reductions in IIS increase lifespan [1–3], yet in the clinic, either an enhanced supply of insulin  or sensitization to this hormone  to alleviate established insulin resistance also improve longevity. In addition, in the last 10 years, results have begun to emerge indicating that many of the risk factors for T2D (Type 2 diabetes), including hyperinsulinaemia and insulin resistance, are also associated with AD (Alzheimer's disease) (see for example ), which has led to growing interest in the possibility of identifying in IIS some common targets for therapy [7,8].
The FOXO (forkhead box O) transcription factor family is recognized as the key downstream effector of IIS longevity signals , regulating much of the pleiotropic action of IIS on cell differentiation, proliferation and metabolism . Previous studies established that IIS induction triggers FOXO1a inhibition and exclusion from the nucleus by PI3K (phosphoinositide 3-kinase)-dependent, PKB (protein kinase B)/Akt- and CK1 (casein kinase 1)-mediated phosphorylation of FOXO1a on five residues, Thr24, Ser256, Ser319, Ser322 and Ser325 [10–13]. These residues and nuclear shuttling are conserved in other FOXOs, except for FOXO6, which lacks residues corresponding to Ser319, Ser322 and Ser325 and is predominantly nuclear even when phosphorylated . Our recent studies have investigated Zn2+-dependent phosphorylation of FOXO1a on these residues in response to tropolone and dithiocarbamate compounds , which was accompanied by repression of PEPCK (phosphoenolpyruvate carboxykinase) and G6Pase (glucose 6-phosphatase) , key regulators of hepatic gluconeogenesis [16–25]. Metal-binding compounds have also been attracting interest as anti-neurodegenerative agents, including the 8-hydroxyquinoline compound CQ (clioquinol). A variety of mechanisms have been proposed to account for the effects of chronic CQ treatment in neurodegenerative disease models, including sequestration of Fe2+/Fe3+ , Cu2+ and Zn2+ , Co2+  or alternatively transport of Cu2+  or Zn2+ , resulting in cellular effects, including increased levels of metalloproteinase , proteasome inhibition [29,32], stimulation of TNFα (tumour necrosis factor α)  and regulation of mitochondrial hydroxylase . Prompted by our interest in the effects of Zn2+-binding compounds on IIS, in the present study we have investigated the possibility that Zn2+-dependent regulation of IIS and gluconeogenesis could contribute to the cellular action of CQ.
CQ was used widely in humans until it was withdrawn following cases of subacute myelo-optic neuropathy in Japan [34,35], but the therapeutic prospects for derivatives of this compound might be limited even further because a variety of other 8-hydroxyquinolines were identified in the 1950s as being diabetogenic through the destruction of pancreatic β-cells (Table 1) [36–38]. In common with the effects on IIS that we report in the present study, the diabetogenic effect of 8-hydroxyquinolines is thought to require interaction with Zn2+  but despite this, several Zn2+-binding 8-hydroxyquinolines, including CQ, are non-diabetogenic [36–38,40]. To establish whether or not Zn2+-dependent IIS induction and diabetogenicity may be dissociated, we have compared the effects on IIS of a variety of 8-hydroxyquinolines and other structures whose diabetogenicity is known.
MATERIALS AND METHODS
8CPT-cAMP [8-(4-chlorophenylthio)-cAMP], aluminium chloride, 5-amino-8-hydroxyquinoline, 5-chloro-7-iodo-8-hydroxyquinoline, chromium(III) chloride, copper sulfate, disulfiram, 4-hydroxyquinaldine, 2-hydroxyquinoline, 8-hydroxyquinoline-5-sulfonic acid, manganese chloride, wortmannin and xanthurenic acid were all obtained from Sigma–Aldrich. The PKB inhibitor Akti, PI-103, PD98059 and rapamycin were from Calbiochem. Zinc acetate was obtained from Riedel de Haen. Dithizone, 8-hydroxyquinoline and quinaldine were from Fluka. The compounds used in the present study were dissolved in DMSO and stored aliquotted (10 μl) at −20°C. Aliquots were discarded after one freeze–thaw cycle. All antibodies were as described previously [12,41], except for anti-pFOXO1a-Ser256, anti-pPKB-Ser473 (Cell Signaling Technology) and an anti-actin antibody (Calbiochem/Merck).
Cell culture and lysis
HEK (human embryonic kidney)-293 and HL1c cells were maintained essentially as described previously [11,41]. HEK-293 cells were used for experiments 4 or 5 days after seeding and DMEM (Dulbecco's modified Eagle's medium)/FBS (fetal bovine serum) was replaced 1 day before the experiment. HL1c cells were used for experiments on day 2 after seeding and were serum-starved on the evening before stimulation. The inhibitors PI-103 and wortmannin were added 1 h before stimulation, Akti was added 10 min before stimulation and both PD98059 and rapamycin were added 30 min before stimulation. Cells were lysed on ice using buffer A [50 mM Tris acetate (pH 7.5), 1% (w/v) Triton X-100, 1 mM EDTA, 1 mM EGTA, 10 mM 2-glycerophosphate, 5 mM sodium pyrophosphate, 50 mM NaF, 1 mM sodium orthovanadate, 1 mM benzamidine, 0.2 mM PMSF and 0.1% 2-mercaptoethanol]. The lysates were centrifuged at 13000 g for 5 min at 4°C, and the supernatants were removed and stored at −80°C until use. In all experiments, cells were serum-starved prior to stimulation for at least 30 min.
RT (reverse transcription)–PCR
RT–PCR assays of insulin-sensitive genes were carried out in HL1c cells, cultured in DMEM containing 10% (v/v) FBS and 1g/l glucose. Serum-starved cells were stimulated with agents for 4 h. RNA was obtained using an RNeasy® Mini kit from Qiagen. RNA was reverse-transcribed to produce first-strand cDNA using SuperScript® II reverse transcriptase (Invitrogen). Briefly, RNA, dNTPs and random primers were heated at 65°C for 5 min and then chilled, prior to incubation with the reverse transcriptase at 25°C for 10 min and 42°C for 50 min, before inactivation by heating at 70°C for 15 min. RT–PCR was performed in a 96-well plate with sequence-specific primers and probes. In each experiment, results were normalized to Dex (dexamethasone)/cAMP where Dex/cAMP=100, and, in each sample, RNA levels were normalized to cyclophilin.
Transient transfection and confocal microscopy
HEK-293 cells were plated in serum-free medium in dishes containing a sterile glass coverslip. On the following day, cells were transfected with a GFP (green fluorescent protein)-tagged wild-type FOXO1a construct which was generated as described previously  using FuGENE 6 Transfection Reagent (Roche). The transfection mixture was added dropwise to the dish and incubated overnight. After each treatment, cells were washed with ice-cold PBS, which was immediately removed and replaced with ice-cold 4% (w/v) paraformaldehyde. Cells were fixed in the dark at room temperature (20°C) for 15 min, then washed with ice-cold PBS. Coverslips were mounted on to microscope slides (VWR) with a small drop of Vectashield (Vector Laboratories) and fixed into place using nail varnish. Cells were imaged using a laser scanning confocal imaging system (Leica TCS SP5). The Leica Application Suite software (Advanced Fluorescence 1.8.2) was used to capture fluorescent images with a ×63 magnification oil-immersion lens and ×4 zoom. The nuclei of HEK-293 cells are large enough to be identified easily without counterstaining. The intensity of the fluorescence in the nuclear and cytoplasmic compartments was quantified using the Leica Application Suite software. A ratio was obtained by dividing the nuclear intensity value by the averaged cytoplasmic value (two regions of interest were quantified for the cytoplasm).
RESULTS AND DISCUSSION
CQ and Zn2+ induce acute IIS, and phosphorylation of FOXO1a and its exit from the nucleus
Previously, we developed four antibodies that are capable of detecting endogenous FOXO1a in lysates to study FOXO1a phosphorylation by PKB/Akt and CK1 in response to insulin [12,41] and other regulators of the pathway [43,44]. One antibody detects FOXO1a regardless of phosphorylation state, whereas the others detect FOXO1a phosphorylated on Thr24, Ser256 and Ser325. PI3K-dependent phosphorylation of Ser325 also reported phosphorylation at two priming sites, Ser319 and Ser322 [12,13]. In the initial experiments, we compared the ability of CQ to induce IIS in the presence of 10 μM Cu2+, Zn2+, Cr3+ and Al3+. We used HEK-293 cells because phosphorylation of endogenous FOXO1a is readily detectable in these cells in response to IGF-1  and other agents [15,44]. Similar to our previous experiments with disulfiram and tropolones , we found that combining CQ with Zn2+ induced IIS after 1 h of treatment, whereas Cu2+ was far less effective and other ions were ineffective (Figure 1A). In time-course treatments we found that 1 h of treatment induced FOXO phosphorylation that approached the maximum (Figure 1B), slower than IGF-1, which induces maximal responses in this cell line within 5 min . In dose-response experiments, a concentration of 10 μM approached the maximum (Figure 1C). Previously, we showed that phosphorylation of FOXO1a triggers its exit from the nucleus [11–13], and in the present study we found that CQ/Zn2+ induced nuclear exclusion similarly to insulin/IGF-1 (Figure 2).
Residues Thr24, Ser256 and Ser319 on the FOXOs lie not only within consensus sequences for phosphorylation by PKB/Akt and SGK (serum- and glucocorticoid-induced protein kinase), but also for phosphorylation by p70 S6 kinase and p90RSK (ribosomal S6 kinase). We found that FOXO1a phosphorylation in response to CQ/Zn2+ was sensitive to two PI3K inhibitors, PI-103 and wortmannin (Figure 3A). Further experiments in which HEK-293 cells were incubated with either rapamycin [which prevents activation of p70 S6 kinase by inhibiting mTOR (mammalian target of rapamycin)] or PD98059 (which prevents the activation of p90RSK) showed that neither of these drugs affected CQ/Zn2+-induced phosphorylation of FOXO1a, whereas a previously identified PKB-specific inhibitor [22,45] did inhibit CQ/Zn-induced FOXO1a phosphorylation comparably with IGF-1 (Figure 3B). The results with these kinase inhibitors are similar to FOXO regulators that we have studied previously [12,15,41,44], suggesting that each one induces FOXO phosphorylation by the PI3K- and PKB-sensitive mechanism used by IGF-1/insulin. Further work will be required to establish the target of CQ/Zn2+-dependent effects but taken together with our previous results using structurally unrelated Zn2+-binding compounds , our work with CQ suggests a Zn2+-dependent target, upstream of PI3K in the IIS cascade. PTP1b (protein tyrosine phosphatase 1b) is directly inhibited by Zn2+ in vitro , but other possible targets include inhibition of PTEN (phosphatase and tensin homologue deleted on chromosome 10), activation of receptor kinases and enhanced recruitment of signalling proteins to the receptor.
Effect of quinoline compounds on FOXO1a phosphorylation: critical role of the co-ordinating group at the 8-position
The effect of CQ and Zn2+ led us to investigate the ability of similar compounds to phosphorylate FOXO1a. Many analogues of 8-hydroxyquinoline (Table 1) were identified in the 1950s as being diabetogenic [36–38] and therefore in addition we tested two structurally unrelated diabetogenic substances, alloxan and dithizone. Alloxan is thought not to bind metals, except when transformed to non-diabetogenic alloxanic acid , but in contrast, the metal-binding properties of dithizone are well known and are thought to contribute to its diabetogenic character . In our screen of these compounds we found that dithizone and two other compounds, 8-hydroxyquinoline and 8-hydroxyquinaldine, strongly induced FOXO1a phosphorylation in a Zn2+-dependent manner (Figure 4A). The availability of analogues of 8-hydroxyquinaldine enabled us to investigate the role of the co-ordinating hydroxy group at the 8-position. In 4-hydroxyquinaldine the co-ordinating group is relocated with respect to 8-hydroxyquinaldine, and in quinaldine it is absent (Table 1). Neither of these compounds was capable of inducing IIS, indicating that the presence of a co-ordinating group at the 8-position is critical for the effect on IIS (Figure 4B). Consistent with this, we found that substitution of the 8-OH group with a non-co-ordinating amino group also generates a compound that is inactive on IIS (Figure 4C). The ability of the co-ordinating hydroxy group at the 8-position to enable induction of IIS was conditional, as two further analogues of 8-hydroxyquinoline, xanthurenic acid and 8-hydroxyquinoline-5-sulfonic acid, were ineffective at inducing IIS (Figure 4A). Xanthurenic acid and 8-hydroxyquinoline-5-sulfonic acid have additional charged functional groups with respect to 8-hydroxyquinoline, and we have previously shown that altering the charge on the tropolones also prevents them from stimulating IIS , providing further evidence that charge is an important determinant of the ability of a Zn2+-binding compound to induce IIS, perhaps by inhibiting entry to the cell or specific cellular compartments.
Effect of CQ on gluconeogenic genes
Next we explored the effect of CQ/Zn2+ and the inactive analogue 8-aminoquinoline on hepatic gluconeogenic gene expression using the cell line HL1c, which we have used previously to measure the expression of these genes . Repression of hepatic gluconeogenesis by reduced expression of PEPCK and G6Pase is recognized as a key aspect of the anti-hyperglycaemic action of insulin [16–25]. Previously, we showed that repression of the gluconeogenic genes PEPCK and G6Pase by the unrelated Zn2+-binding small molecules disulfiram and β-thujaplicin mirrored the regulation of FOXO transcription factors by these agents . In the present study we found that CQ/Zn2+ also inhibited PEPCK and G6Pase (Figures 5A and 5B). Consistent with the effects on IIS, neither Zn2+ nor CQ was capable of repressing these genes on their own (Figure 5A). In dose-response experiments in the presence of 10 μM Zn2+, CQ inhibited PEPCK and G6Pase with an IC50 of 4 μM (Figure 5B), which is identical with the effect of disulfiram and Zn2+ . Similar experiments carried out with 8-aminoquinoline showed that this compound was ineffective on gluconeogenic genes (Figure 5C).
CQ dissociates FOXO1a regulation from diabetogenic properties
Comparing our results from the present study with previous ones on diabetogenicity [36–38], we found that, among 8-hydroxyquinoline analogues and dithizone, compounds capable of forming uncharged complexes with Zn2+ induce IIS and tended to be diabetogenic, whereas those that do not affect IIS because they are charged or do not bind Zn2+ tended to be non-diabetogenic (Table 1). This suggests that the ability to access the cell and to bind Zn2+ is important for diabetogenicity and the ability to induce IIS. It seems unlikely, however, that diabetogenicity is causally linked to IIS-dependent FOXO inhibition because in pancreatic β-cells , in common with other tissues such as muscle, IIS induction and FOXO inhibition maintain cell mass and promote cell survival [48,49]. Moreover, in the case of CQ, which protects mice from streptozotocin-induced diabetes , IIS induction is dissociated from diabetogenicity (Table 1). Previous studies suggested that 8-hydroxyquinoline acidifies and destroys β-cell insulin-secretory granules by liberating protons following binding of Zn2+  but with CQ, acidification may be reduced because the positioning of the electron-withdrawing halogens results in a much lower pKa value than in 8-hydroxyquinoline. Proton release is not required for Zn2+-dependent IIS induction because we have recently found that another Zn2+-binding substance DEDTC (diethyldithiocarbamate) , which is known to protect against dithizone-induced diabetes , and which does not liberate protons on interaction with Zn2+ in cells , induces IIS at least as potently as dithizone or 8-hydroxyquinoline . Another effect of halogenation is to render CQ more hydrophobic than other 8-hydroxyquinolines, but no more so than dithizone which is also diabetogenic, suggesting that increased hydrophobicity alone is unlikely to account for the lack of diabetogenicity of CQ. Taken together, the results of the present study suggest that there may be scope to design small metal-binding FOXO1a regulators that are non-diabetogenic, particularly by targeting hydrophobic structures that do not liberate protons on interaction with Zn2+. It will also be interesting to determine the role of IIS in the opposing effects of streptozotocin and CQ not only in diabetes , but also in neurodegeneration, where evidence suggests that intracerebral streptozotocin administration results in pathology resembling AD .
In the present study we have investigated the ability of 8-hydroxyquinoline and related compounds to induce IIS and repress gluconeogenic genes. We find that several compounds based on 8-hydroxyquinoline, including CQ, an experimental therapy for AD, are capable of inducing IIS and regulating gluconeogenic genes in a strictly Zn2+-dependent manner. This ability to induce IIS does not tolerate absence or relocation of the co-ordinating group at the 8-position, and additional functional groups that change the charge of the molecule also prevented IIS induction, consistent with the notion that the effects are metal- rather than ligand-dependent. Many 8-hydroxyquinolines have previously been reported to be diabetogenic, but our present results with CQ and previously with DEDTC suggest that it is feasible to identify non-diabetogenic Zn2+-dependent IIS-inducing agents. CQ has been investigated in experimental models of AD, Parkinson's disease and Huntington's disease, and it will be interesting to determine whether or not the IIS responses that we have studied, including FOXO regulation and gluconeogenic gene expression, contribute to the effects of the drug in these disease settings in vivo.
Amy Cameron carried out most of the blotting and microscopy, and all of the gene expression, experiments. Additional blots were performed by Katherine Wallace, Jean Harthill and Graham Rena. Additional help with microscopy was provided by Lisa Logie, Alan Prescott and Terry Unterman. The paper was written by Graham Rena and then improved in the light of comments from the other authors.
This work was supported by the Anonymous Trust, Diabetes UK [grant number 08/0003666], the Medical Research Council, the Rank Prize Funds and Tenovus Scotland.
We thank Professor Mike Ashford and Dr Calum Sutherland for helpful discussions on the project. We thank Dr Craig Beall for help with using the microscope and associated sample preparation and analysis.
Abbreviations: AD, Alzheimer's disease; CK1, casein kinase 1; CQ, clioquinol; DEDTC, diethyldithiocarbamate; Dex, dexamethasone; DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; FOXO, forkhead box O; GFP, green fluorescent protein; G6Pase, glucose-6-phosphatase; HEK, human embryonic kidney; IGF, insulin-like growth factor; IIS, insulin/IGF-1 signalling; PEPCK, phosphoenolpyruvate carboxykinase; PI3K, phosphoinositide 3-kinase; PKB, protein kinase B; RSK, ribosomal S6 kinase; RT, reverse transcription; T2D, Type 2 diabetes
- © The Authors Journal compilation © 2012 Biochemical Society